Structure of Gramicidin A in a Lipid Bilayer Environment Determined

Andrew H. Beaven, Alexander J. Sodt, Richard W. Pastor, Roger E. Koeppe, II, ..... D Peter Tieleman, Justin L MacCallum, Walter L Ash, Christian Kandt...
0 downloads 0 Views 1MB Size
jgp

S u p p le m ental m aterial

THE JOURNAL OF GENERAL PHYSIOLOGY

Jensen et al., http://www.jgp.org/cgi/content/full/jgp.201210820/DC1

Simulation and analysis details Simulations were performed on a special-purpose machine, Anton, designed for MD simulations (Shaw et al., 2009) in the NPT ensemble (310 K, 1 bar, Berendsen coupling scheme with one temperature group; Berendsen et al., 1984). All bond lengths to hydrogen atoms were constrained using M-SHAKE (Kräutler et al., 2001). Van der Waals and short-range electrostatic interactions were cut off at 10 Å. Long-range electrostatic interactions were calculated using GSE (Shan et. al., 2005) with a 64 × 64 × 64 FFT mesh. The simulation time step was 2 fs (2.5 fs in simulations 14 and 15); long-range electrostatics were evaluated every third step. The protein was initially relaxed in the membrane before a constant electric field was applied (Table S1). The field was increased linearly to full strength over a time period of 20 Å, while including only simulation snapshots where three ions were bound inside the SF and no ions were present in the cavity.

Figure S6.  Water and lipid head-group orientation

in gA. The water dipole moment (z component only) was calculated from simulations with no ions in the system and at different applied voltages (simulations 17–19);  is the angle between the (unit) water dipole vector and the z axis. (Inset; bottom left) The z component of the pore–water dipole moment per molecule (in units of 2.35 D) is graphed as a function of time. (Inset; top right) The distribution of the lipid head group P–N vector tilt angle is graphed relative to the z axis. The lipid head-group orientation was found to be independent of the applied voltage used in the simulation. The orientation of the single-file water molecules inside the pore (|z| < 20 Å) differed with voltage. At 0 mV, the water molecules aligned with their dipoles either parallel or antiparallel to the z axis, and P1 on average equaled zero; they reoriented (concertedly) with a rate of 3.6 µs1. With the field applied along +z, the propensity of water molecules to align with their dipoles parallel to the field increased; P1 was thus greater than zero and increased with the magnitude of the applied field. The reduced net water orientation inside the pore at lower voltages increased the pore electrostatic potential (Fig. 4 G), opposing ion permeation. 

Jensen et al.

S3

Figure S7.  gA (1MAG) deuterium quadrupole splittings (DQS) of tryptophan (Trp) residues 9 and 11 at different applied voltages. The data indicate that the average orientation of these two Trp residues is independent of the applied field. Consequently, voltage-dependent conformational changes of these two residues appear not to influence ion permeation; their contribution to the pore electrostatic potential in Fig. 4 G in the main text is similar at all voltages. See the Fig. S3 legend for more details.

Figure S8.  gA (1MAG) tryptophan (Trp) rotameric states at different applied voltages. These data indicate, in accordance with Fig. S7, that the average orientation of the Trp residues is independent of the applied field. Consequently, voltage-dependent reorientation of the Trp residues appears not to influence ion permeation. See the Fig. S4 legend for more details.

S4

Atomistic simulation of single-file ion conduction

Table S1 KV1.2/2.1 simulations Simulation

t

E



V

NK

NK,cav

IK

IK,cav

IW

IW/IK

O

mwt

res/O

mwt res/O

1

18

0.2000

86.8 ± 0.2 752.9 ± 1.4

4,196 8,162 21.12 ± 2.83

40.56 ± 73.71 14.40 ± 0.60

0.68 ± 0.10

2.54

0.8 ± 0.1

0.4 ± 0.0

0.4 ± 0.1

2

14

0.1750

86.5 ± 0.2 656.4 ± 1.2

1,472 1,463 16.47 ± 1.30

16.28 ± 1.19

0.57 ± 0.06

2.68

1.0 ± 0.1

1.0 ± 0.0

0.0 ± 0.1

1,373 1,403 13.37 ± 0.77

3

17

0.1500

86.4 ± 0.1 561.9 ± 0.9

4

10

0.1375

86.6 ± 0.1 516.4 ± 0.5

599

647

9.67 ± 1.10

9.40 ± 0.63

13.59 ± 0.73

6.74 ± 0.63

0.50 ± 0.06

2.67

1.2 ± 0.1

1.2 ± 0.0

0.0 ± 0.1

10.29 ± 1.11

3.97 ± 0.58

0.41 ± 0.08

2.87

1.7 ± 0.2

1.6 ± 0.0

0.1 ± 0.2

5

10

0.1250

86.9 ± 0.1 471.2 ± 0.5

375

515

6.05 ± 0.63

8.19 ± 0.47

3.25 ± 0.38

0.54 ± 0.08

2.78

2.7 ± 0.3

2.5 ± 0.1

0.1 ± 0.3

6

15

0.1250

86.9 ± 0.1 471.1 ± 0.6

567

764

5.91 ± 0.69

7.89 ± 0.50

3.15 ± 0.33

0.53 ± 0.08

2.76

2.7 ± 0.3

2.6 ± 0.1

0.1 ± 0.3

7

44

0.1000

86.7 ± 0.2 375.8 ± 0.8

358

604

4.16 ± 0.42

7.00 ± 0.32

1.63 ± 0.15

0.39 ± 0.05

2.90

3.8 ± 0.4

3.6 ± 0.1

0.3 ± 0.4

8

22

0.0750

86.6 ± 0.1 281.6 ± 0.4

152

386

1.77 ± 0.47

4.79 ± 0.88

0.76 ± 0.13

0.43 ± 0.14

2.45

9.2 ± 2.4

7.9 ± 0.6

1.2 ± 2.4

36.3 ± 9.0

31.2 ± 3.5

5.1 ± 9.7

9

22

0.0500

87.1 ± 0.1 188.9 ± 0.2

60

520

0.44 ± 0.11

3.72 ± 0.32

0.48 ± 0.15

1.09 ± 0.44

2.44

10

40

0.0250

87.6 ± 0.1

94.9 ± 0.1

11

575

0.05 ± 0.04

2.30 ± 0.18

0.55 ± 0.12

1.91 ± 0.97

2.08 310.4 ± 213.9 112.6 ± 30.9 197.8 ± 216.1

87.3 ± 0.3

47.3 ± 0.1

20

11

22

0.0125

331

0.17 ± 0.04

2.36 ± 0.64

0.16 ± 0.08

0.97 ± 0.52

2.33

96.7 ± 25.4

74.9 ± 8.9

21.8 ± 26.9

12 a

29

0.1000 130.9 ± 0.2 567.5 ± 1.0

1,820 2,031

9.98 ± 0.63

10.69 ± 0.55

6.09 ± 0.70

0.61 ± 0.08

2.58

1.6 ± 0.1

1.6 ± 0.0

0.0 ± 0.1

13b

18

0.1000 152.4 ± 0.2 661.0 ± 1.0

1,847 1,784 16.41 ± 1.31

15.69 ± 1.49

27.39 ± 13.80 1.67 ± 0.85

2.44

1.0 ± 0.1

1.0 ± 0.0

0.0 ± 0.1

14c

16

0.060

144.7 ± 0.4 376.5 ± 1.0

237

568

2.58 ± 0.33

6.15 ± 0.52













15c

15

0.060

151.4 ± 0.1 393.9 ± 0.3

168

431

1.72 ± 0.51

4.55 ± 0.32













16d

23

0.1000

87.8 ± 0.1 380.6 ± 0.4

318 1,020

2.20 ± 0.17

6.99 ± 12.63

1.68 ± 0.24

2.62

7.3 ± 0.6

7.2 ± 0.3

0.113 ± 0.6

17d

13

0.1000 130.8 ± 0.1 567.2 ± 0.8

531

604

6.53 ± 0.48

4.40 ± 0.49

0.67 ± 0.09

2.51

2.5 ± 0.2

1.0 ± 0.0

1.432 ± 0.2

18e,f

6

0.1500

87.0 ± 0.2 565.9 ± 1.0

196

194 10.59 ± 2.6



8.22 ± 2.89

0.78 ± 0.33

0.23

1.5 ± 0.4

11.9 ± 1.1

10.4 ± 1.2

19f

28

0.0500

87.0 ± 0.2 188.6 ± 0.4

28

119

0.34 ± 0.10



0.08 ± 0.03

0.23 ± 0.11

1.70

47.1 ± 13.8

30.0 ± 4.8

16.8 ± 14.6

20f

35

0.0250

87.4 ± 0.2

94.7 ± 0.2

9

33

0.18 ± 0.07



0.08 ± 0.02

0.44 ± 0.21

1.99

88.9 ± 34.6

133.0 ± 54.5 44.1 ± 64.6

21f

18

0.0125

87.0 ± 0.2

47.2 ± 0.1

6

46

0.09 ± 0.04



0.07 ± 0.04

0.78 ± 0.56

1.45 177.8 ± 79.0

69.0 ± 41.8 108.8 ± 89.4

22g

9(4)

0.075

80.9 ± 0.3 263.0 ± 1.0

8

44

0.30 ± 0.09















23g

10

0.175

81.2 ± 0.2 616.0 ± 1.8

205

222

3.23 ± 0.81















24g

9(8)

0.075

81.3 ± 0.3 264.4 ± 0.9

18

120

0.47 ± 0.23















25g

12(6)

0.150

81.4 ± 0.3 529.4 ± 1.6

68

92

1.68 ± 0.39















27.96 ± 3.64

0.76 ± 0.13

Total simulated time (t, µs), applied field (E; kcal · mol1 · Å1 · e1), average box length (; Å), applied voltage (V = E · < lz>), K+ permeation events across pore and cavity (NK, NK,cav), K+ pore and cavity currents and water currents (IK, IK,cav, IW; pA), ion-to-water permeation ratio (IW/IK), K+ kinetic occupancy (O), mean waiting time between pore permeation events (mwt; 108 s), K+ SF residence time in sites S0–S5 (res/O; occupancy normalized, summed over all sites, 108 s; see Figs. 2, C and E, and S1), mean waiting time and residence time difference (mwtres/O; 108 s). The time in parentheses is the simulated time used for analysis in the three simulations where the pore underwent half (simulations 22 and 25) or full (simulation 23) pore closure because of the absent VSDs. a n = 158k atoms. b n = 186k atoms. c n = 230k atoms with T1 retained; simulation 15 used CHARMM27 lipid parameters. d [KCl] = 0.3 M. e The SF deteriorated after 5 µs. f With  (K+–O) = 3.2874 Å. g Pore-only “controls” in a palmitoyl oleoyl phosphatidylethanolamine membrane, as used previously in our pore-only simulations, with the force field– corrected SF. Simulations 22 and 23 used CHARMM36 lipid parameters (Klauda et al., 2010), whereas simulations 24 and 25 used the CHARMM27 lipid parameters (MacKerell et al., 1998) we used previously for POPE (Jensen et al., 2010).



Jensen et al.

S5

Table S2 gA simulations t

E



V

 (K+–O)

1

77

0.06

68.8

183.7

2

45

0.1

68.8

306.3

3

29

0.15

68.8

4

28

0.2

5

11

6 7

Simulation

NK

IK

O

3.0859

7

0.02 ± 0.01

1.00

3.0859

14

0.06 ± 0.02

1.00

459.4

3.0859

40

0.22 ± 0.05

1.00

68.9

613.5

3.0859

93

0.55 ± 0.09

1.00

0.27

69.0

829.1

3.0859

107

1.61 ± 0.37

1.00

21

0.2

72.6

645.8

2.9399

5

0.06 ± 0.01

1.00

30

0.2

72.6

645.8

3.0000

26

0.16 ± 0.04

1.07a

8

35

0.2

72.6

645.8

3.1000

88

0.41 ± 0.07

1.00

9

31

0.2

72.5

645.6

3.1500

80

0.41 ± 0.08

1.00

10

30

0.2

72.5

645.6

3.2000

49

0.25 ± 0.05

1.01

11

5

0.2

72.5

645.6

3.2874

0

0

1.00b

12c

11

0.2

72.6

645.8

3.0000, 3.2000

0

0

1.00

13c

5

0.2

72.6

645.8

3.0859, 3.2000

0

0

1.00

14c

16

0.2

72.5

645.6

3.2000, 3.0000

47

0.47 ± 0.08

1.01

15c

11

0.2

72.5

645.6

3.0859, 3.0000

57

0.85 ± 0.15

1.00

16c

44

0.1

72.4

322.1

3.0859, 3.0000

22

0.08 ± 0.03

1.00

17a

5

0

77.2

0

3.0859







18a

7

0.06

77.4

201.2

3.0859







19a

7

0.2

77.5

671.7

3.0859







20d

10

0.15

68.8

459.4

3.0859

12

0.21 ± 0.07

1.00

21e

20

0.2

75.1

670.0

3.0859

10

0.11 ± 0.03

1.01

22e

11

0.27

74.0

890.0

3.0859

30

0.44 ± 0.14

1.01

23f

5

0

86.6

0

3.0859







24f

7

0.06

86.5

225.0

3.0859







25f

7

0.2

85.7

742.0

3.0859







Total simulated time (t; µs), applied field (E; kcal · mol · Å · e ), average box length (; Å), applied voltage calculated as V = E · , LJ K+–O interaction parameter (; Å), K+ permeation events (NK), current (IK; pA), and kinetic occupancy (O) of the pore. a With no ions in the system. 10% of the configurations with a K+ populated pore had two ions in the pore. b Only 87 occurrences were observed with the ion being inside the pore at 15 < z < 15 Å. c Different  values were used for vestibule and lumen. d Used the 1JNO structure (Townsley et al., 2001). e Using CHARMM27 lipid parameters. f Using CHARMM27 lipid parameters and with no ions in the system. 1

1

1

Table S3 gA permeation kinetics Simulation

V

NK

NK,vestibule

IK

IK,vestibule

IK/IK,vestibule

mwt

res/O

mwtres/O

5

829.1

4

613.5

107

96

1.61 ± 0.37

1.43 ± 0.25

1.13 ± 0.11

0.99

0.95

0.04

93

123

0.55 ± 0.09

0.73 ± 0.07

0.75 ± 0.02

2.89

1.99

3

0.9

459.4

40

75

0.22 ± 0.05

0.42 ± 0.07

0.52 ± 0.02

7.21

3.12

4.1

2

306.3

14

132

0.06 ± 0.02

0.47 ± 0.10

0.13 ± 0.00

27.03

2.31

24.7

1

183.7

7

240

0.02 ± 0.01

0.50 ± 0.05

0.04 ± 0.00

90.11

0.54

89.6

+

Applied voltage (V; mV), K permeation events across pore and vestibule (NK, NK,vestibule; a buffer region of 2 Å was used in vestibule calculations), pore and vestibule currents (IK, IK,vestibule; pA). The latter comprises current across minima 1 and 3 in Fig. 4 D in the main text, their ratio (IK/IK,vestibule), mean waiting time between permeation events (MWT; 107 s), cumulative K+ pore residence time (res/O; summed over minima 1–10 in Fig. 4 D and occupancy normalized, O = 1; 107 s), and difference between mean waiting and residence times (entry time, mwt  res/O; 107 s).

S6

Atomistic simulation of single-file ion conduction

Table S4 K+ transition ratios in gA Simulation

t

E

V

 (K+–O)

IK (pA)

1→2

2→3

3→4

4→5

2→3/1→2 3→4/2→3 4→5/3→4

1

77

0.06

183.7

3.0859

0.02 ± 0.01

14,795

815

153

43

0.1

0.2

0.3

2

45

0.10

306.3

3.0859

0.06 ± 0.02

8,692

837

211

35

0.1

0.3

0.2

3

29

0.15

459.4

3.0859

0.22 ± 0.05

5,199

893

288

128

0.2

0.3

0.4

4

28

0.20

613.5

3.0859

0.55 ± 0.09

4,292

1,171

506

174

0.3

0.4

0.4

5

11

0.27

829.1

3.0859

1.61 ± 0.37

1,310

546

251

144

0.4

0.5

0.6

6

21

0.20

645.8

2.9399

0.06 ± 0.01

2,056

466

164

30

0.2

0.4

0.2 —

11

5

0.20

645.6

3.2874

0

165

3

0

1

0.0

0

12a

11

0.20

645.8

3.0000, 3.2000

0

3,429

11

0

0

0.0

0



13a

5

0.20

645.8

3.0859, 3.2000

0

997

96

10

2

0.1

0.1

0.2

14a

16

0.20

645.6

3.2000, 3.0000

0.47 ± 0.08

1,376

79

31

48

0.1

0.4

1.6

15a

11

0.20

645.6

3.0859, 3.0000

0.85 ± 0.15

1,564

363

425

184

0.2

1.1

0.4

16a

44

0.10

322.1

3.0859, 3.0000

0.08 ± 0.03

8,213

1,150

650

171

0.1

0.6

0.3

+

Total simulated time (t; µs), applied field (E; kcal · mol · Å · e ), applied voltage (V; mV), LJ K –O interaction parameter (; Å), and current (IK; pA). For an idealized linear I-V curve, the individual transition ratios would remain constant across all voltages, whereas the absolute number of transitions would be proportional to the magnitude of voltage. The percentage of K+ transitions from site 2 to site 3 relative to transitions from site 1 to 2—the 2→3/1→2 ratio (see Fig. 4 D in the main text)—differs about fourfold between lowest and highest voltages. In contrast, corresponding variation of the 4→5/3→4 ratio is only about twofold. a Dual  values were used for vestibule and lumen. 1

1

1

Table S5 KcsA simulations with native and weakened SF–K+ interactions Simulation

t

E



V

NK

Native

2

0.10

96.3

428.5

257

Weakened

2

0.10

96.1

427.4

19

Asp80 “up”

2

0.10

96.1

427.5

Ala71Glu

2

0.15

95.6

Asp80Asn

2

0.15

95.0

IK

IW

IW/IK

O

25.59 ± 2.43

0.73 ± 0.42

0.03 ± 0.02

3.25

1.63 ± 0.44

0.77 ± 0.36

0.47 ± 0.26

2.16

188

19.11 ± 1.91

0

0

2.83

638.1

200

21.74 ± 3.53

9.64 ± 1.48

0.44 ± 0.10

2.91

634.1

168

15.65 ± 6.51

3.52 ± 1.64

0.23 ± 0.14

3.33

The Asp80 1 side chain was in one simulation restrained to pointing into the extracellular solution (“up,” as the corresponding residue is predominantly oriented in KV1.2/2.1) using dihedral restraints with a force constant of 9 kcal · mol1 · rad2. Two simulations, one with Glutamate at position 71 (“Ala71Glu”) and one with an Asp80Asn mutant (“Asp80Asn”) were carried out to examine the effects of introducing protonated Glu in the SF region and of removing the Asp80 negative charge, respectively. For each simulation the following are listed: the total simulated time (t; µs), applied field (E; kcal · mol1 · Å1 · e1), average box length (; Å), applied voltage calculated as V = E · < lz>, K+ permeation events (NK), K+ and water currents (IK, IW; pA), ion-to-water permeation ratio (IW/IK), and K+ kinetic occupancy (O).



Jensen et al.

S7